Text summarization based on sentiment classification of comments from online Vietnamese newspaper

Abstract

To know opinions of consumers regarding products or public about important problems in society, then the best and most effective way is to exploit information of community from Internet and social network. Today is an era of information explosion through Internet and social networking, so we are able to exploit effectively information from the huge sources. The opinion of individuals is not only objective information but also contains emotions of the author. It through Internet has big power to make a stream of public opinion that will impact on network community. This is really an enormous subjective information resource, then it will have great meaning for many areas, such as economics, politics, society and culture if we have methods and techniques to exploit it effectively. An automatic system classifying comments based on sentiment is really necessary to exploit efficiently this resource. In order to support users have more concise and appropriate information, then question of summary information should be studied and solved, especially on side of the views and sentiments of each opinion. To exploit the resource effectively to summary information, the paper will propose a text Vietnamese summary model, not only based on semantics but also based on sentiment features. We have built a base model to solve this problem. We have exploited and developted methods summarizing and sentiment analysing for our proposed model. Our system can draw Vietnamese comments from online Vietnamese newspaper, analyze the sentiments of comments, classify them and make a summary of opinions effectively.